Advances in the Application of 18F-FDG PET/CT Radiomics for Diagnosis, Treatment and Prognosis Prediction of Lymphoma

CHENG Ran,HU Jiajia,LI Biao

DOI: https://doi.org/10.3969/j.issn.1674-8115.2023.06.016

2023-01-01

Abstract:Lymphoma is a highly heterogeneous hematological malignancy that can affect multiple organs throughout the body, exhibiting significant clinical variations among its subtypes. 18F-fluorodeoxyglucose (18F-FDG) PET/CT plays a crucial role in the clinical diagnosis and treatment of lymphoma by facilitating anatomical localization and quantification of metabolic characteristics of highly aggressive lymphomas. This imaging examination method enables a comprehensive evaluation by comparing the metabolic changes before and after treatment, as well as the metabolic difference between lesions and blood pools. However, the heterogeneity of lymphoma, coupled with the limitations of 18F-FDG PET/CT in differentiation, poses challenges for physicians and adversely impacts the clinical treatment plan and prognosis of patients. With the advancement of computer hardware and image analysis technology, radiomics technology, based on the extraction of imaging features of lesions for analysis and diagnosis, has emerged. Numerous researchers have dedicated their efforts to exploring imageomics in lymphoma assessment by using 18F-FDG PET/CT. By integrating feature data with relevant clinical information, models have been developed to effectively correlate image information, clinical data, pathology, and survival outcomes, thereby enhancing the accuracy and efficiency of imaging diagnosis. Furthermore, the utilization of predictive models for prognosis and treatment efficacy has the potential to mitigate subjective errors arising from disparities in physician experience, thereby contributing to the realization of personalized medicine. This review intends to comprehensively summarize the research progress of 18F-FDG PET/CT radiomics in the diagnosis, treatment and evaluation of lymphoma in recent years, from the aspects of diagnosis and differential diagnosis, prognosis prediction and risk grading, drug efficacy prediction and radiomics analysis algorithm optimization, so as to provide insights for future research in machine learning and the development of medical imaging analysis techniques.